Center-led procurement: the evidence gap behind the dominant operating model

Sixty-three percent of enterprises over $2 billion in procurement spend use a center-led or hybrid operating model. It is the dominant structural choice for large, multi-business organizations. Consulting firms and procurement advisors have recommended it for over a decade. Companies that adopt it describe the logic with confidence: centralize strategy and leverage, decentralize execution and responsiveness. The best of both worlds.

Here is what that same cohort cannot do: isolate whether changing the operating model actually improved procurement outcomes.

The consensus that built an industry standard

The center-led model makes intuitive sense. Centralize category strategies, supplier relationships, technology, and analytics under one procurement center of excellence (COE). Push operational buying to business units within defined guardrails. Procurement gets the leverage of a consolidated spend base without the friction of a headquarters team that does not understand local needs.

Multiple consulting frameworks have reinforced this logic. The model emerged as the natural evolution from the centralized-versus-decentralized debate that dominated procurement in the 1990s and 2000s. Centralized was too rigid. Decentralized leaked too much value. Center-led promised to fix both problems at once. By the early 2020s, it had become the default recommendation for any organization above a certain size and complexity threshold.

McKinsey's research shows procurement functions managing 50% more spend per full-time equivalent than five years ago, with functions becoming 25-40% more efficient. A global insurance company increased strategic headcount by 20%, created a COE with more than ten new skills, and doubled spend under procurement's influence. These are real results from organizations that restructured. The question is not whether restructuring coincided with improvement — it is whether the restructuring itself caused it.

The isolation problem: what actually changed when you restructured

Organizations rarely change only their operating model. A typical restructuring initiative includes at least three to five simultaneous changes: a new CPO or procurement leadership, a technology platform upgrade, renegotiated supplier contracts, revised KPIs, and a headcount restructure that moves people between roles. When results improve, any of these variables could be the driver. Attribution is impossible without a controlled comparison — which no organization runs on itself.

This is not unique to procurement. Organizational design suffers from the same attribution problem everywhere. Marketing restructures during a brand campaign. Sales reorganizes during a product launch. Procurement restructures during a technology implementation or a market cycle shift. The timing is rarely coincidental — a new CPO is hired to drive change, and restructuring is part of the change package. But the package obscures the model's independent contribution.

• New CPO hired → operating model change is part of their mandate
• Technology upgrade (P2P, CLM, analytics) → efficiency gains are attributed to the model, not the tool
• Market cycle shift → commodity prices drop, savings increase, model gets credit
• Headcount restructure → team composition changes, harder to separate from structural change
• Supplier renegotiation wave → savings attributed to better category management, not just better timing

McKinsey's own data illustrates the ambiguity. Procurement functions that "facilitate the best quality, engagement, and cost performance" achieved EBITDA margin improvements of five percentage points or more. These functions tend to share certain characteristics — strong COEs, upgraded talent, digital tools, clear governance. But the research does not isolate the operating model as the causal variable. It identifies correlated factors. Correlation is not attribution.

The complexity cost nobody measures

Center-led models introduce structural complexity. Category managers report to a central procurement leader for strategy, methodology, and performance while serving business unit stakeholders who control budget, demand, and day-to-day priorities. The "two masters" problem is well-documented: when the central procurement function pushes for cost reduction through supplier consolidation and the business unit demands speed through local sourcing, the category manager faces conflicting directives without clear decision rights.

Hackett Group's 2023 Key Issues research found that 67% of procurement professionals said they needed formal governance models between procurement and business stakeholders — a finding that suggests the governance question in hybrid models is unresolved, not settled. The center-led model requires clear decision rights to function. Most organizations implement the structure without defining who wins when central and local priorities conflict.

There are real costs to this ambiguity. Slower decision cycles when escalation is required. Category managers spending disproportionate time on stakeholder management instead of category work. Savings targets that central procurement sets but cannot enforce because business units control actual buying behavior. The model's theoretical efficiency can be consumed by its practical coordination cost — and few organizations measure this trade-off.

What separates the models that work from the ones that drift

Despite the evidence gap, some center-led implementations produce clear, measurable improvements. The ones that work share four characteristics that the ones that drift do not.

Three questions to ask before your next restructuring conversation

What variable are we actually testing? If you are changing the operating model, technology platform, leadership, and KPIs in the same initiative, you cannot attribute results to the model. Pick one variable to isolate. Run the technology upgrade first. Measure the baseline. Then restructure. The sequencing produces data; the bundle produces narrative. Deliverable: a single-variable change plan with pre- and post-measurement windows. Timeline: 90 days of baseline data before restructuring begins.

Who wins when central and local priorities conflict? If the answer is "it depends" without a clear decision-rights framework, your center-led model has governance debt. Write the escalation paths before you write the org chart. Specify which categories are centrally mandated, which are locally owned, and what triggers an override. Deliverable: a decision-rights charter covering the top 20 spend categories. Timeline: 30 days.

How will we know if the model is working? Define specific, measurable outcomes tied to the model change — not to concurrent initiatives. If technology is upgrading at the same time, strip out technology-driven savings from the model's scorecard. If market conditions are shifting, normalize for the cycle. The metric is not "did results improve" — it is "did the model contribute to improvement that would not have occurred under the previous structure." Deliverable: a model-specific scorecard with three to five metrics isolated from concurrent variables. Timeline: before the restructuring launch.